Ixxiv THE VOYAGE OF H.M.S. CHALLENGER. 



stem-forms, since they retain the original spherical shape (Pis. 5-8, 1 1-30). In the other 

 three suborders a vertical main axis is developed, which in P r u n o i d e a is longer, in 

 D i s c o i d e a shorter than the other axes of the shell. Hence the shell of the P r u n- 

 oidea (p. 284, Pis. 13, his, 17, 39,40) is ellipsoidal or cylindrical, that of the 

 Disco idea, on the other hand, lenticular or discoidal (p. 402, Pis. 31-38, 41-48). 

 Finally, the shell of the fourth suborder, L a r c o i d e a, is lentelliptical ; it has the 

 ground-form of a triaxial ellipsoid, and is characterised by the possession of three 

 unequal dimensive axes, or three isopolar axes of different lengths perpendicular to each 

 other (p. 599, Pis. 9, 10, 49, 50). 



110. The Sheletpn of the Acantharia. — -The skeletons of the Acantharia. or Actipy- 

 LEA are distinguished from those of all other Radiolaria by two very important peculiarities ; 

 in the first place, they consist not of silica but of a peculiar organic substance, Acanthin, 

 and secondly, their development is centrogenous, numerous radial spines or acanthin 

 spicules being formed which are united in the middle of the central capsule. Hence the 

 Acantharia are the only Radiolaria in which the skeleton originates from the first in 

 the middle of the central capsule. The number of radial spines is primitively indefinite, 

 variable, and often considerable (more than a hundred), but in the great majority it is 

 limited to twenty. In accordance with this the legion may be divided into two orders, 

 the more archaic small group Adekicantha, with an indefinite number of spines, and the 

 more recent group, Icosacautha, which has been developed from them and possesses 

 twenty regularly disposed spines ; of the three hundred and seventy-two species of Acan- 

 THAJiiA which have been hitherto described, about five per cent, belong to the former, about 

 ninety-five per cent, to the latter division (see note A, below). The numerous genera of 

 Icosacantha may then l^e again divided into two suborders, of which the Acanthonida 

 (p. 740, Pis. 130-132) produce no complete lattice-shell, and thus agree with the 

 Actinelida, with which they may be united as Acanthometra in the broader 

 sense (or Acantharia without a lattice-shell). The A c a n t h o p h r a c t a, on the other 

 hand (p. 791, Pis. 133-140), produce a complete lattice-shell, usually by means of 

 two opposite or four crossed transverse processes, which arise from each radial sjaine and 

 unite with each other (see note B, below). In most Acanthophracta the lattice-shell 

 remains single ; only in the Phractopeltida does it consist of two concentric lattice-spheres 

 (p. 847, PI. 133, figs. 1-6). Furthermore, the whole order Acanthophracta may 

 be subdivided into two suborders according to the different ground-form of the lattice- 

 shell; this remains spherical in the Sphserophracta (the three families Sphserocapsida, 

 Dorataspida, Phractopeltida, Pis. 133-138). On the other hand, it assumes another 

 form in the Prunophracta; it becomes ellipsoidal in the Belonaspida (PI. 136, 

 figs. 6-9), discoidal or lentiform in the Hexalaspida (PI. 139); and finally takes the 

 shape of a double cone in the Diploconida (PI. 140). 



